Project 2

#loading all the necessary libraries
library(gutenbergr)
library(tidytext)
library(tidyverse)
library(readtext)
library(sentimentr)
library(ggiraphExtra)
library(ggplot2)
library(RColorBrewer)
library(scales)
library(plotly)
library(htmlwidgets)

#finding texts by Fitzgerald on gutenberg
fitzgerald_text <- gutenberg_works(author == "Fitzgerald, F. Scott (Francis Scott)")
fitzgerald_text <- fitzgerald_text[-c(2,3), ]

#downloading all the texts found
fitzgerald_download <- fitzgerald_text %>% 
                      gutenberg_download(meta_fields = c("title", "author"))

Determining mirror for Project Gutenberg from http://www.gutenberg.org/robot/harvest
Using mirror http://aleph.gutenberg.org

The_Great_Gatsby <- read_delim("The Great Gatsby_Fitzgerald, F. Scott (Francis Scott)_clean.txt", 
    delim = ".", quote = "\\\"", escape_double = FALSE)

Rows: 41 Columns: 1
── Column specification ───────────────────────────────────────────────────────────────
Delimiter: "."
chr (1): text

ℹ Use `spec()` to retrieve the full column specification for this data.
ℹ Specify the column types or set `show_col_types = FALSE` to quiet this message.

The_Great_Gatsby <- The_Great_Gatsby %>% mutate(author = "Fitzgerald, F. Scott (Francis Scott)", .after = text) %>% mutate(title = "The Great Gatsby", .before = author)

Tender_is_the_Night <- read_delim("Tender is the Night_Fitzgerald, F. Scott (Francis Scott)_clean.txt", 
    delim = ".", quote = "\\\"", escape_double = FALSE)

Rows: 3346 Columns: 1
── Column specification ───────────────────────────────────────────────────────────────
Delimiter: "."
chr (1): text

ℹ Use `spec()` to retrieve the full column specification for this data.
ℹ Specify the column types or set `show_col_types = FALSE` to quiet this message.

Tender_is_the_Night <- Tender_is_the_Night %>% mutate(author = "Fitzgerald, F. Scott (Francis Scott)", .after = text) %>% mutate(title = "Tender is the Night", .before = author)

fitzgerald_download$gutenberg_id <- NULL
fitzgerald_corpus <- bind_rows(fitzgerald_download, The_Great_Gatsby, Tender_is_the_Night)

Warning: One or more parsing issues, see `problems()` for details
Warning: One or more parsing issues, see `problems()` for details

The_Sun_Also_Rises_Hemingway_Ernest <- read_delim("The Sun Also Rises_Hemingway, Ernest_clean.txt", 
    delim = "\t", quote = "\\\"", escape_double = FALSE, 
    trim_ws = TRUE)
The_Sun_Also_Rises_Hemingway_Ernest <- The_Sun_Also_Rises_Hemingway_Ernest %>% mutate(author = "Hemingway, Earnest", .after = text) %>% mutate(title = "The Sun Also Rises", .before = author)

The_Old_Man_and_the_Sea_Hemingway_Ernest <- read_delim("The Old Man and the Sea_Hemingway, Ernest_clean.txt", 
    delim = ".", quote = "\\\"", escape_double = FALSE)
The_Old_Man_and_the_Sea_Hemingway_Ernest <- The_Old_Man_and_the_Sea_Hemingway_Ernest %>% mutate(author = "Hemingway, Earnest", .after = text) %>% mutate(title = "The Old Man and the Sea", .before = author)

For_Whom_the_Bell_Tolls_Hemingway_Ernest <- read_delim("For Whom the Bell Tolls_Hemingway, Ernest_clean.txt", 
    delim = ".", quote = "\\\"", escape_double = FALSE)
For_Whom_the_Bell_Tolls_Hemingway_Ernest <- For_Whom_the_Bell_Tolls_Hemingway_Ernest %>% mutate(author = "Hemingway, Earnest", .after = text) %>% mutate(title = "For Whom The Bell Tolls", .before = author)

A_Farewell_to_Arms_Hemingway_Ernest <- read_delim("A Farewell to Arms_Hemingway, Ernest_clean.txt", 
    delim = ".", quote = "\\\"", escape_double = FALSE)
A_Farewell_to_Arms_Hemingway_Ernest <- A_Farewell_to_Arms_Hemingway_Ernest %>% mutate(author = "Hemingway, Earnest", .after = text) %>% mutate(title = "A Farewell to Arms", .before = author)

The_Garden_of_Eden_Hemingway <- read_delim("The Garden of Eden_Hemingway_clean.txt", 
    delim = "\t", quote = "\\\"", escape_double = FALSE, 
    trim_ws = TRUE)
The_Garden_of_Eden_Hemingway_Ernest <- The_Garden_of_Eden_Hemingway %>% mutate(author = "Hemingway, Earnest", .after = text) %>% mutate(title = "The Garden of Eden", .before = author)

hemingway_corpus <- bind_rows(The_Sun_Also_Rises_Hemingway_Ernest,The_Old_Man_and_the_Sea_Hemingway_Ernest,For_Whom_the_Bell_Tolls_Hemingway_Ernest,A_Farewell_to_Arms_Hemingway_Ernest,The_Garden_of_Eden_Hemingway_Ernest)

all_authors <- bind_rows(fitzgerald_corpus, hemingway_corpus)

tidy_all_authors <- all_authors %>% unnest_tokens(word, text) %>% anti_join(stop_words)

Joining, by = "word"

#breaking up texts into words
tidy_hemingway <- hemingway_corpus %>% unnest_tokens(word, text) %>% anti_join(stop_words)

Joining, by = "word"

tidy_all_authors <- all_authors %>% unnest_tokens(word, text) %>% anti_join(stop_words)

Joining, by = "word"

nrc_list <- get_sentiments("nrc")
counter <- tidy_hemingway %>% inner_join(nrc_list) %>% count(word, sort = TRUE)

Joining, by = "word"

sentiment_abs_h <- tidy_hemingway %>% inner_join(get_sentiments("nrc") %>% 
                 filter(sentiment %in% c("anger", 
                                         "anticipation", "disgust", "fear", "joy", "sadness", "surprise", "trust"))
    ) %>%
  count(sentiment) %>%
  pivot_wider(names_from = sentiment,
              values_from = n,
              values_fill = 0) %>% mutate(author = "Hemingway")#%>% 

Joining, by = "word"

  #mutate(sentiment = positive - negative)
total_words_hemingway <- NROW(tidy_hemingway)
sentiment_rel_h <- sentiment_abs_h[,-9]/total_words_hemingway
sentiment_rel_h <- sentiment_rel_h*100
sentiment_rel_h <- sentiment_rel_h %>% mutate(author = "Hemingway")
#anger  anticipation    disgust fear    joy sadness surprise    trust

nrc_joy <- get_sentiments("nrc")
#tidy_hemingway %>% inner_join(nrc_list) %>% count(word, sort = TRUE)
sentiment_abs_f <- tidy_fitzgerald %>% inner_join(get_sentiments("nrc") %>% 
                 filter(sentiment %in% c("anger", 
                                         "anticipation", "disgust", "fear", "joy", "sadness", "surprise", "trust"))
    ) %>%
  count(sentiment) %>%
  pivot_wider(names_from = sentiment,
              values_from = n,
              values_fill = 0) %>% mutate(author = "Fitzgerald")

Joining, by = "word"

total_words_fitzgerald <- NROW(tidy_fitzgerald)
sentiment_rel_f <- sentiment_abs_f[,-9]/total_words_fitzgerald
sentiment_rel_f <- sentiment_rel_f*100
sentiment_rel_f <- sentiment_rel_f %>% mutate(author = "Fitzgerald")

sentiment_both <- bind_rows(sentiment_rel_h, sentiment_rel_f)

data_vis_2 <- ggRadar(
  data = sentiment_both,
  mapping = aes(color = author),
  rescale = FALSE,
  interactive = TRUE,
  size = 2,
  grid.min = 1,
  centre.y = 0,
  legend.position = "right"
)
data_vis_2

emotion_list <- read_csv("emotion_list.txt")

Rows: 486 Columns: 1
── Column specification ───────────────────────────────────────────────────────────────
Delimiter: ","
chr (1): word

ℹ Use `spec()` to retrieve the full column specification for this data.
ℹ Specify the column types or set `show_col_types = FALSE` to quiet this message.

emotion_words <- tibble(emotion_list, emotion = TRUE)

hemingway_emotions <- tidy_hemingway %>% left_join(emotion_words)

Joining, by = "word"

fitzgerald_emotions <- tidy_fitzgerald %>% left_join(emotion_words)

Joining, by = "word"

#counting how often female words appeared
#table shows each text and the respective percentage
fitzgerald_emotion_calc <- fitzgerald_emotions %>% group_by(title) %>% count(emotion) %>% mutate(percent = round(n/sum(n)*100, 2)) %>% mutate(author = "F. Scott Fitzgerald")

#counting how often female words appeared
#table shows each text and the respective percentage
hemingway_emotion_calc <- hemingway_emotions %>% group_by(title) %>% count(emotion) %>% mutate(percent = round(n/sum(n)*100, 2)) %>% mutate(author = "Earnest Hemingway")

combined_table <- bind_rows(hemingway_emotion_calc, fitzgerald_emotion_calc) %>% filter(emotion == TRUE)

data_vis_1 <-ggplot(data=combined_table, aes(x=percent, y=title, fill=author, text = paste("% of emotion: ", percent, sep = ""))) +
  geom_bar(stat="identity") + labs(x = "% of emotion detected", 
       y = "All novels", fill = "Author", title = "Emotive content in all novels")
       
data_vis_1_int <- ggplotly(data_vis_1, tooltip = "text")
data_vis_1_int

save(author_table, hemingway_corpus, fitzgerald_corpus, data_vis_1_int, data_vis_2, file = "saved_data.rda")

puretext <- bind_rows(tidy_hemingway[, 3], tidy_fitzgerald[, 3])
purestring <- unlist(puretext)
sad_1 <- paste(unique(str_subset(purestring, "^anger|^angr|^disgust|^contempt|^revuls|^envy|^envi|^jealous|^exaspera|^frustra|^aggrava|^agitat|^annoy|^grouch|^grump|^irrita|^bitter|^dislike|^feroci|^fury|^furi|^hate|^hostil|^loath|^outrage|^rage|^rag|^resent|^scorn|^spite|^venge|^wrath|^torment|^fear|^horr|^alarm|^fright|^hyster|^mortifi|^panic|^shock|^terror|^nervous|^anxi|^apprehen|^distress|^dread|^tense|^uneasi|^uneasy|^worr|^joy|^cheer|^amuse|^bliss|^delight|^ecsta|^elati|^enjoy|^euphor|^gaiety|^glad|^glee|^happi|^jolli|^jolly|^jovial|^jubil|^satisf|^content|^pleasur|^enthrall|^raptur|^eager|^hope|^optimis|^pride|^proud|^triumph|^relie|^zest|^enthusias|^excit|^exhilara|^thrill|^zeal|^love|^ador|^affec|^attrac|^care|^cari|^compassion|^fond|^like|^liking|^sentiment|^tender|^long|^arous|^desire|^infatuat|^lust|^passion|^sad|^disappoint|^dismay|^displeasur|^alien|^defeat|^deject|^embarrass|^homesick|^humiliat|^insecur|^isola|^insult|^lonel|^neglect|^reject|^depress|^despair|^gloom|^glum|^grie|^hopeless|^melancholi|^miser|^sorrow|^unhappi|^woe|^shame|^guilt|^regret|^remorse|^suffer|^agon|^anguish|^hurt|^pity|^piti|^sympath|^surprise|^amaze|^astonish")), collapse = ' ')